Published Paper Details:

E-commerce platforms facilitate customers in posting reviews for a wide range of products. Typically, a single product can accumulate numerous lengthy and repetitive reviews, making it challenging for customers to gather relevant information for informed decision-making. Therefore, the application of automatic review summarization holds immense potential in assisting customers by delivering concise and genuine summaries of online reviews on E-commerce websites. In this study, we propose a Fine-Tuning Pre-Trained Model, which produces more precise summaries closely resembling those generated by humans. Additionally, our system evaluates the overall sentiment of the generated summaries, enabling customers to quickly grasp the underlying tone of the text. For sentiment analysis, a supervised learning model using the Naive Bayes classifier has been trained on social media sentiment analysis data, achieving an accuracy exceeding 98%. The abstractive summarization model adopts a Sequence-to-Sequence approach with an attention mechanism, resulting in reasonably good review summaries.